One of the most common complaints about AEO is that it is hard to measure. There is no AI Search Console, no official citation report, and no single dashboard that shows you how often ChatGPT or Claude is mentioning your brand. But the measurement gap is smaller than it looks. AI referral traffic is trackable in GA4, citation rates are testable with a simple manual protocol, and AEO readiness scores give you a reliable leading indicator of where citation rates are heading.
This guide covers the complete measurement framework: what to track, how to track it, and how to interpret the signals.
Layer 1: AI referral traffic in GA4
GA4 rolled out a native 'AI Assistant' default channel through May and June 2026, reaching broad availability across properties on 7 June 2026. When a session arrives with a referrer from a recognised AI chatbot, GA4 now assigns it the medium 'ai-assistant' and groups it under this new channel automatically, with nothing to configure. Out of the box it recognises ChatGPT, Gemini, and Claude.
The native channel is a genuine improvement, but it has two limits worth knowing before you rely on it. First, it only counts traffic from launch onward, so there is no historical backfill and no clean year-on-year comparison yet. Second, a large share of AI-driven visits, often well over a third, arrive with no referrer at all (mostly from mobile apps) and still land in Direct rather than AI Assistant, so the native channel undercounts true AI referral volume.
To cover both gaps, keep a manual segment running alongside the native channel. In GA4, create a segment or custom channel group rule filtering sessions where the session source contains the following domains:
- chat.openai.com (ChatGPT web)
- chatgpt.com (ChatGPT direct)
- perplexity.ai (Perplexity)
- claude.ai (Claude web)
- bard.google.com / gemini.google.com (Gemini)
- copilot.microsoft.com (Copilot)
- you.com (You.com AI search)
In GA4: Reports > Acquisition > Traffic Acquisition. Set the dimension to 'Session source'. Filter by the domains above. This gives you a direct count of sessions driven by AI engine citation clicks, independent of whether the native channel has classified them. For most sites, this number will be small today but growing 15-30% month over month. Track it monthly, alongside the native AI Assistant channel figure.
For a channel group that survives the native rollout's lack of history, add a custom rule above Referral in your GA4 channel group settings: Medium equals referral, Source matches a regex of your AI domains (perplexity\.ai|chatgpt\.com|claude\.ai|gemini\.google\.com|copilot\.microsoft\.com). Move the rule above Referral so AI sessions are bucketed there instead of falling through to generic referral traffic, giving you one figure that spans before and after the native channel existed.
Layer 2: Manual citation testing
Referral traffic only captures users who click through from AI answers. Many AI citations drive brand awareness without a click. Manual testing captures the full picture.
Set up a monthly citation audit with this protocol:
- Identify your 10-15 highest-value queries: the questions your target customers most commonly ask in your category
- Submit each query to ChatGPT, Claude, Perplexity, and Gemini in a fresh session
- Record: Was your brand mentioned? Was it cited with a link? What position (first, second, third)? Which competitors were cited instead?
- Track in a spreadsheet: query, engine, cited (yes/no), position, competitor citations
- Calculate your citation rate: (queries where you were cited / total queries) per engine
This protocol takes 30-45 minutes per month and gives you the most direct measure of AI visibility available. Run it on the same date each month to control for query drift and model update effects.
Layer 3: Brand mention monitoring
Not all AI mentions are citations with links. In some engines (especially ChatGPT for queries within training data), your brand may be mentioned by name in the answer body without a linked citation. These mentions still drive brand recognition and intent.
To track brand mentions in AI answers, extend your manual testing protocol to record not just citations with links but any mention of your brand name or product name in the answer text. Also set up Google Alerts for your brand name to catch any web-published coverage that includes your brand in an AI context.
Layer 4: AEO readiness score as a leading indicator
Referral traffic and citation testing are lagging indicators: they tell you what happened. AEO readiness scores are leading indicators: they tell you where citation rates are heading.
An AEO score reflects the technical and content signals that drive citation likelihood. When your score improves on the crawler access check, schema check, or E-E-A-T check, citation rate improvements typically follow within 4-8 weeks as crawlers re-index your pages and retrieval indices refresh.
Run a CiteRank audit at baseline, again after each major implementation sprint, and then quarterly. Track your score on each of the 10 checks, not just the overall score. A check that moves from failing to passing is a citation rate improvement in progress. If you are still comparing options, see our checklist for picking an AEO checking tool.
Building your AEO reporting dashboard
A complete AEO reporting dashboard covers four metrics, updated monthly:
- AI referral sessions (from GA4, count and month-over-month change)
- Citation rate per engine (from manual testing: % of target queries where you are cited)
- Brand mention rate (from manual testing: % of queries where your brand name appears in the answer)
- AEO readiness score (from CiteRank: overall score and per-check breakdown)
The correlation between AEO score and citation rate is not instantaneous, but it is reliable. Sites that consistently improve their AEO score over a 90-day period see corresponding improvements in citation rate in the following 90 days. Use this correlation to build the business case for AEO investment.
The fastest way to get your baseline metrics: run a CiteRank audit (5 seconds, free) to get your readiness score, then run the 10-query manual test in ChatGPT and Perplexity this afternoon. You will have your full baseline in under an hour.
What good AEO metrics look like
Benchmarks vary significantly by industry, but for a typical B2B SaaS site with 6-12 months of AEO investment:
- AEO readiness score: 65+ is the threshold where citation rates become consistent
- Citation rate in Perplexity (easiest engine to get cited by): 30-50% of target queries
- Citation rate in ChatGPT: 15-30% of target queries for well-optimised sites
- Citation rate in Claude: 10-25% of target queries (higher bar due to E-E-A-T weighting)
- AI referral sessions: growing 15-30% month over month is healthy for a site actively investing in AEO